文献类型：期刊文献

英文题名：Portable and Versatile Electronic Nose System Based on Edge Computing and Multi-task Model

作者：Ni, Wangze[1,2]; Wang, Tao[3]; Zhu, Jiaqing[4]; Li, Zhuoheng[1,2]; Chen, Lechen[1,2]; Cheng, Weiwei[4]; Mei, Haixia[5]; Xuan, Fuzhen[3]; Yang, Jianhua[1,2]; Zeng, Min[1]; Hu, Nantao[1,2]; Yang, Zhi[1]

机构：[1] Shanghai Jiao Tong University, National Key Laboratory of Advanced Micro and Nano Manufacture Technology, Shanghai, 200240, China; [2] Shanghai Jiao Tong University, Department of Micro/Nano Electronics, School of Electronic Information and Electrical Engineering, Shanghai, 200240, China; [3] School of Mechanical and Power Engineering, East China University of Science and Technology, Shanghai Key Laboratory of Intelligent Sensing and Detection Technology, Shanghai, 200237, China; [4] School of Materials Science and Engineering, Shanghai University of Engineering Science, Shanghai, 201620, China; [5] Changchun University, Key Lab Intelligent Rehabil & Barrier free Disable, Changchun, 130022, China

年份：2024

外文期刊名：Proceedings of IEEE Sensors

收录：EI(收录号：20250317719661)

语种：英文

外文关键词：Deep learning - Electronic health record - Multi-task learning

摘要：The electronic nose (E-nose) has been widely used in various gas detection scenarios, such as environmental monitoring and smart homes. However, the lack of portability, versatility, and low predictive accuracy of existing E-nose systems have significantly hindered their further application across different fields. This study proposes a portable and versatile E-nose system to address these issues. A plug-in connection between the two boards is achieved by designing the system to separate the sensor board from the main board, enabling fast sensor replacement. An artificial neural network (ANN) is deployed on the microcontroller unit of the main board. Once gas response data is obtained from the sensor board, the onboard ANN model analyzes the data and simultaneously predicts the type and concentration of the target gas independently. Notably, the E-nose system can achieve high accuracy in multi-task deep learning with a prediction time of only 2 ms. Benefiting from edge computing and the multi-task model, our E-nose system is expected to be utilized in high-performance and real-time air quality monitoring. ? 2024 IEEE.